X-ray microtomography analysis of soil pore structure dynamics under wetting and drying cycles

Highlights • µCT allowed quantifying morphological changes in the region of the sample close to the hydraulic contact.• 3D images permitted detailed analysis of the pore shape and size distribution.• Tortuosity and pore connectivity was affected by wetting and drying cycles.• Soil water retention curve was influenced by wetting and drying cycles.


INTRODUCTION
friction, clay dispersion, aggregate size and stability can be induced by the 70 application of W-D (Rajaram and Erbach, 1999). 71 Thus, possible changes in soil pore structure in different regions of the soil 72 sample could help to explain differences in water retention curve when samples   in the city of Ponta Grossa, PR, Brazil (25°06'S, 50°10'W, 875 m above sea level). 120 The soil was an Oxisol (Rhodic Hapludox) according to USDA soil taxonomy (Soil 121 Survey Staff, 2013). The soil was classified as a clay texture with 17% sand, 30% 122 silt and 53% clay. The particle density and the amount of C content evaluated 123 were 2.41 g cm -3 and 60.7 g kg -1 , respectively. cylinders for this study was due to their use for water retention curve 133 measurements. Since the soil water content is very important at the sampling 134 time, to minimize damage in the soil structure, samples were collected when soils 135 were near their field capacity, about three days after a high intensity rainfall event.  137 analysis 138 Soil samples were saturated by the capillary rise method. The wetting (W) 139 procedure consisted in soaking the samples in a tray with the water level just 140 below the top of the steel cylinders. This procedure was kept over a period of 2 141 days to allow saturation of the sample and to avoid the presence of the entrapped 142 air bubbles, which can cause slaking of soil aggregates (Klute, 1986 152 The wetting procedure to saturate the samples was exactly the same as 153 that described in the previous section. Following the saturation, the samples were 154 placed in contact with the porous media (sand) on the suction table. The samples 155 were equilibrated in the pressure heads varying from -10 to -100 cm of H2O with 156 intervals of 10 cm (Romano et al., 2002). After the thermodynamic equilibrium 157 was reached (nearly 4-5 days for each sample) the moist soil mass was evaluated 158 using a precision balance (0.01 g). The dry soil mass was obtained at the end of 159 the water retention curve by oven drying for 48 h at 105 °C.

160
The experimental pairs of data obtained (soil water contents and pressure 161 heads) were fitted using the mathematical model proposed by van Genuchten-

162
Mualem equation (van Genuchten, 1980). The Excel solver based on the total 163 sum of squares was used for fitting the experimental data. The soil water retention 164 curve adjustments were obtained using the average values of soil water contents 165 (n=6). In order to check the quality of the water retention curve fitting, the root-166 mean-square error and the coefficient of determination were calculated. Relative 167 differences (RD) were also obtained between the water retention curves in order 168 to evaluate the effect of the different W-D on the soil pore structure.   The original grey-level X-ray microtomographic images were processed 206 using ImageJ 1.42 software (Rasband, 2007). An unsharp mask procedure with 207 1 voxel standard deviation and weighing 0.8 was applied to enhance the edge 208 contrast. The segmentation process was based on the nonparametric and 209 unsupervised Otsu method for thresholding (Otsu, 1979). The remove outlier tool      were observed (Fig. 2b). We also noticed that the number of pores did not differ 283 between ROIW and ROIHC for all W-D cycles analyzed. However, soil pore 284 structure changes as shown by the porosity increase were not influenced by the 285 increase in the number of pores after W-D cycles mainly for ROIHC (Table 1).

286
The application of W-D cycles can provoke swelling and shrinkage 287 processes in the soil volume, which cause tension forces between aggregates.

288
The action of these forces can reduce soil porosity when the force is directed   The increase in the porosity influenced positively the number of junctions 359 and branches and negatively the pore connectivity for ROIW and ROIHC (Table 1). The increase in pore connectivity was accompanied by a decrease in the 365 number of junctions and branches mainly for ROIHC, which was more susceptible 366 to changes in relation to the whole sample (Table 1)  The decrease in the average tortuosity was followed by an increase in pore 377 connectivity and in the number of junctions and branches for ROIW and ROIHC 378 (Table 3). These results indicate that more aligned pores were characterized by 379 a greater number of connected pores, mainly for ROIHC. This is interesting 380 because these two morphological properties are known to influence water   390 The distribution of pore sizes was affected by the W-D cycles for ROIW and 391 ROIHC (Fig. 4). Volume of pores presented a significant decrease between ROIs analyzed the same behavior was noticed between 0, 6 and 12 W-D cycles, 395 except for 12 W-D cycles for pores with sizes between 0.1 and 1 mm 3 (Fig. 4c).

396
The influence of these pore classes in soil porosity was greater for ROIHC in 397 comparison to ROIW (Figs. 4a, b). 398 For the largest pores (>10 mm 3 ), the ROIW volume of pores was 399 significantly larger than that of ROIHC for 0, 6 and 12 W-D cycles (Fig. 4e). Volume 400 of pores also increased with the application of W-D cycles for the largest pore 401 sizes for ROIHC and ROIW. This result explains the increase in soil porosity (Table   402 1), which is related to an increase in the number of pores joined together.  (Table 1), we reinforce the importance of these changes mainly when occurring 425 in the region close to the bottom of the sample (ROIHC).

426
The distribution of pores in terms of shape presented differences between 427 ROIs with the W-D cycles for the equant and triaxial shaped pores (Fig. 5). The 428 cycles caused an increase in the equant shaped volume of pores for ROIW, while 429 the opposite was observed for ROIHC (Fig. 5a). For the triaxial shaped pores, a 430 decrease in these pore types was recorded for ROIW (Fig. 5d). Few significant 431 correlations were measured between the distribution of pores in terms of shape 432 and the micromorphological properties investigated, mainly for ROIHC (Table 1).

433
For ROIW, pore shape was related with pore connectivity and tortuosity in the x   446 The water retention curves showed the W-D cycles treatment influence in 447 the soil structure. In our study computed tomography was used to reveal the  (Table 2). Higher water retention was found for the range of pressure heads 457 analyzed with 12 W-D cycles (Fig. 6a). This implies that the application of 12 W-458 D caused an increase in pores from to textural to structural pore size ranges (from 459 21 30 to >100 µm equivalent cylindrical diameter), i.e., medium to coarse pores. This 460 result is partially supported by the computed tomography data (Fig. 4). The 461 application of W-D cycles can promote changes in fine matrix pores especially 462 when clayey soils are dried due to the susceptibility of the soil to swelling and 463 shrinkage. W-D cycles can also cause changes in the largest pores as in our 464 study which helps to explain the results for the water retention for 12 W-D cycles 465 (Fig. 4e).

466
Differences of around 10% were recorded between -5 to -100 cm when 467 the samples were submitted to 12 W-D in relation to 0 W-D cycles (Fig. 6b).  The largest difference between 0 and 6 W-D cycles was around 2% for -478 20 to -100 cm (Fig. 6b). This means that only after the application of more than 6 479 W-D cycles, the soil under zero-tillage presented important modifications to its 480 structure. This was confirmed by the comparison between 6 and 12 W-D cycles.